Subj : Software defects
To   : James Byrnes
From : Vince Coen
Date : Tue Jan 29 2002 03:59 am

Hello James!

Monday January 28 2002, James Byrnes writes to All:

JB> I was reading an article about buggy software in InfoWeek the other
JB> day which stated that even an experienced programer introduced one
JB> defect in every ten lines of code.  It went on to say that if 99% were
JB> caught in testing that would still be 1000 errors in a million line
JB> app.

JB> I'm just hobbyist programmer but 1 in 10 seems awfully high to me.
JB> What do you real programmers think?

The figure would vary depending on language used but 1:10 is wrong. 3to5 in 100
is more possible but that includes all errors.
Basic programmer testing at Unit and Link would reduce that down to below
1:500. System, Integration and UAT testing would reduce that down to below
1:15000.

Again subject to quality of testing. Which would be subject to budget allowance
for testing allocated by the Project Manager which is normally 30-45%.

Vince

--- Maximus v3.01/GoldED/2 2.50+#10UK3 under OS/2
* Origin: Air Applewood; OS/2 Gateway to Essex +44-1279-792300 (2:257/609)